The basic concept of personalized medicine is to tailor the treatment for a patient based on his or her genetic makeup, clinical conditions and other personal characteristics to improve efficacy and safety. The coming of the big data era enables us to characterize individual in fine pictures and make the "personalized" clinical decision truly personalized. Such an approach has great potential for improving disease prevention, diagnosis and treatment. For example, in a typical randomized clinical trial aiming for proving the efficacy of a treatment, the final conclusion is drawn based on the average treatment effect in the entire study population. It is possible that while the average treatment effect is near null, the treatment may still be beneficial to a subgroup of patients whose identification prior to the treatment is thus very important. The overall objective of statistical analysis in this area is to provide a data-based empirical estimator for the personalized treatment effect, which can be used to identify subgroup of patients who may benefit the most from a treatment. In this study, we first propose to develop robust statistical methods for estimating the group-specific treatment effect. The proposed approach incorporating many existing methods as special cases depends on minimum model assumptions and provides a general framework for generalization and improvement. We will also discuss how to use the estimated personalized treatment effect to stratify patient population into clinically meaningful strata for better assisting the decision making of clinicians. Secondly, we will study a regularize principal components analysis method for dimension reduction in structured high-dimensional data. The output from the analysis can be used to summarize the characteristics of individual patient as well as for predicting future clinical outcomes of interest. Multiple methods can be used to estimate the treatment effect and form the corresponding treatment selection strategy. Therefore it is important to evaluate and compare the performance of such strategies. Thus our last aim is to develop a systematic robust procedure for evaluating the performance of the personalized treatment effect estimation and associated treatment selection strategy.

Public Health Relevance

For a given treatment, the effect may be very different for different patients. Therefore it is important to develop methods predicting the personalized treatment effect and discovering subgroup of patients who may benefit from a treatment. In this proposal, we plan to study the statistical methods to help achieve these important goals by analyzing empirical data.

Agency
National Institute of Health (NIH)
Type
Research Project (R01)
Project #
2R01HL089778-05
Application #
8695864
Study Section
Biostatistical Methods and Research Design Study Section (BMRD)
Program Officer
Wolz, Michael
Project Start
Project End
Budget Start
Budget End
Support Year
5
Fiscal Year
2014
Total Cost
Indirect Cost
Name
Stanford University
Department
Miscellaneous
Type
Schools of Medicine
DUNS #
City
Stanford
State
CA
Country
United States
Zip Code
94304
Payne, Rebecca; Neykov, Matey; Jensen, Majken Karoline et al. (2016) Kernel machine testing for risk prediction with stratified case cohort studies. Biometrics 72:372-81
Zhao, Lihui; Claggett, Brian; Tian, Lu et al. (2016) On the restricted mean survival time curve in survival analysis. Biometrics 72:215-21
Payne, Rebecca; Yang, Ming; Zheng, Yingye et al. (2016) Robust risk prediction with biomarkers under two-phase stratified cohort design. Biometrics 72:1037-1045
Li, Junlong; Zhao, Lihui; Tian, Lu et al. (2016) A predictive enrichment procedure to identify potential responders to a new therapy for randomized, comparative controlled clinical studies. Biometrics 72:877-87
Shen, Yuanyuan; Cai, Tianxi; Chen, Yu et al. (2015) Retrospective likelihood-based methods for analyzing case-cohort genetic association studies. Biometrics 71:960-8
Minnier, Jessica; Yuan, Ming; Liu, Jun S et al. (2015) Risk Classification with an Adaptive Naive Bayes Kernel Machine Model. J Am Stat Assoc 110:393-404
Uno, Hajime; Tian, Lu; Claggett, Brian et al. (2015) A versatile test for equality of two survival functions based on weighted differences of Kaplan-Meier curves. Stat Med 34:3680-95
Claggett, Brian; Tian, Lu; Castagno, Davide et al. (2015) Treatment selections using risk-benefit profiles based on data from comparative randomized clinical trials with multiple endpoints. Biostatistics 16:60-72
Shen, Yuanyuan; Liao, Katherine P; Cai, Tianxi (2015) Sparse kernel machine regression for ordinal outcomes. Biometrics 71:63-70
Matsouaka, Roland A; Li, Junlong; Cai, Tianxi (2014) Evaluating marker-guided treatment selection strategies. Biometrics 70:489-99

Showing the most recent 10 out of 33 publications